appear, until finally there would remain a single one of much, increased serviceableness (Figs. 18 and 19).
3. For the reasons already given we may suppose the stamens to have their anthers so modified as to open by hinged valves,[1] while at the same time there was developed upon each filament a pair of nectar glands (Fig. 18). Insect visitors, finding an abundance of nectar in a flower, would be less likely to feed upon the pollen, which is so precious to the plant. As the position 
Fig. 19.—Diagram of euberberis flower. Bracts and sepals as before; petals, six, with nectar glands; stamens, six, with valves but no glands; carpel, one; ovules, few. of the nectar is nearer the center of the flower, the visitor comes to occupy a more definite place relative to the pistil and stamens. The six stamens of the inner row are for the most part the only ones which can have their anthers touched, for, as will be seen from the diagram, the remainder are so placed as to be directly behind the others. Being thus superfluous as pollen-producers, the anthers of the other stamens would naturally degenerate, and if they follow the general rule of stamens in flowers which are provided with an abundance of building material (as, for example, the "double" flowers of the florists), they would change into something very like petals. If these petaloid organs became slightly arched over the inner stamens, they might still be of use in the floral household by giving better protection to the pollen than it had previously had, and at the same time increase somewhat the conspicuousness of the blossom. While it is by no means clear that any advantage is gained by having such an organ bilobed at the upper end, it might be a not unnatural result of that special part's having been derived from a bilobed anther. A glance at Figs. 3 and 16 will show that just such a petaloid organ is situated behind each of the stamens in a barberry flower,[2]
- ↑ If it be supposed that the flowers were originally erect, it is possible that this peculiar modification of the dehiscence may have arisen as a protection against rain, which would thus be hindered from washing away the pollen, or indeed quite prevented from so doing if the valve could have had that power of closing in wet weather and opening in dry which Kerner ascribes (Pflanzenleben, p. 123) to the anther valves of certain Lauraceæ. At the present day, as we have seen, the barberry stamens are so well shielded from the rain by the pendent attitude of the flowers that any such peculiarities of the anthers can hardly be of much service in this particular. Still, the assumption that this was equally true in the ancestral forms is of course unwarranted.
- ↑ In B. vulgaris it is the rule for these petals to be entire, its near relative, B. canadensis, having them bilobed. Fig. 3 and also Fig. 16 were, however, drawn directly from undoubted specimens of B. vulgaris.
